Parameter control methods and systems for improving the efficiency of sludge granulation treatment

Abstract

This invention discloses a parameter control method and system for improving the efficiency of sludge granulation treatment, belonging to the field of intelligent control technology. The method includes: extracting a sludge treatment process unit set from the BIOCOS sludge process system, consisting of a P-tank unit, a B-tank unit, and a SU-tank unit; deploying a multimodal monitoring sensor group on this unit set to collect multimodal process unit data; establishing a sludge granulation assessment channel to evaluate the data and obtain sludge granulation state parameters; constructing a control strategy library by combining the unit set and the sludge granulation treatment target; determining the control parameters through optimization and analysis to achieve closed-loop control of sludge granulation. This invention solves the technical problem in existing technologies where the lack of accurate multimodal monitoring and sludge granulation state assessment leads to insufficient efficiency and stability in sludge granulation treatment, achieving closed-loop precise control of sludge granulation treatment and improving the efficiency and stability of sludge granulation treatment.

Description

Technical Field

[0001] This invention relates to the field of intelligent control technology, and in particular to a parameter control method and system for improving the efficiency of sludge granulation treatment. Background Technology

[0002] Control and regulation systems are crucial for ensuring stable operation and improving efficiency in fields such as industrial production and environmental management. Existing technologies often achieve parameter regulation through traditional control and regulation methods. However, as the complexity of application scenarios increases, traditional technologies are gradually revealing their limitations, such as insufficient control precision and delayed regulation response, resulting in poor system management and control effects and failing to meet the needs of efficient and precise regulation. Summary of the Invention

[0003] This application provides a parameter control method and system for improving the efficiency of sludge granulation treatment, which solves the technical problem that the existing technology lacks accurate multimodal monitoring and sludge granulation status assessment, resulting in insufficient efficiency and stability of sludge granulation treatment.

[0004] The first aspect of this application provides a parameter control method for improving the efficiency of sludge granulation treatment. The method includes: extracting key process flows from the BIOCOS sludge process system to obtain a set of sludge treatment process units, the set consisting of a sequence of P-tank units, B-tank units, and SU-tank units; sequentially deploying a multimodal monitoring sensor group on the sludge treatment process unit set to collect sludge treatment multimodal process unit data; establishing a sludge granulation assessment channel, evaluating the sludge state based on the sludge granulation assessment channel to obtain sludge granulation state parameters; constructing a sludge process unit control strategy library based on the sludge treatment process unit set and the sludge granulation treatment target, performing control optimization analysis on the sludge granulation state parameters based on the sludge process unit control strategy library to determine sludge process unit control parameters, and performing closed-loop control of sludge granulation through the sludge process unit control parameters.

[0005] A second aspect of this application provides a parameter control system for improving the efficiency of sludge granulation treatment. The system includes: a sludge treatment process unit set acquisition module, used to extract key process flows from the BIOCOS sludge process system to obtain a sludge treatment process unit set, wherein the sludge treatment process unit set consists of a sequence of P-tank units, B-tank units, and SU-tank units; and a multimodal process unit data acquisition module, used to sequentially deploy a multimodal monitoring sensor group on the sludge treatment process unit set, and collect sludge treatment multimodal process unit data through the multimodal monitoring sensor group; sludge granulation state... The sludge state parameter acquisition module is used to build a sludge granulation assessment channel, and to perform sludge state assessment on the sludge treatment multimodal process unit data based on the sludge granulation assessment channel to obtain sludge granulation state parameters. The sludge process unit control parameter acquisition module is used to construct a sludge process unit control strategy library according to the sludge treatment process unit set and sludge granulation treatment target, to perform control optimization analysis on the sludge granulation state parameters based on the sludge process unit control strategy library, to determine the sludge process unit control parameters, and to perform sludge granulation closed-loop control through the sludge process unit control parameters.

[0006] One or more technical solutions provided in this application have at least the following technical effects or advantages: This application extracts a set of sludge treatment process units consisting of P tank, B tank, and SU tank from the BIOCOS sludge process system. A multimodal monitoring sensor group is deployed on these unit sets. The collected multimodal process unit data is processed through a sludge granulation assessment channel to obtain sludge granulation state parameters. A sludge process unit control strategy library is constructed, and the state parameters are analyzed and optimized using this strategy library to determine the process unit control parameters and implement closed-loop control. This improves the sludge granulation treatment efficiency, stabilizes the sludge granulation effect, and ensures precise and controllable processing. It achieves closed-loop precise control of sludge granulation treatment, thus improving the efficiency and stability of sludge granulation treatment. Attached Figure Description

[0007] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0008] Figure 1 This is a schematic flowchart of a parameter control method for improving the efficiency of sludge granulation treatment provided in an embodiment of this application.

[0009] Figure 2This is a schematic diagram of the parameter control system for improving the efficiency of sludge granulation treatment provided in the embodiments of this application.

[0010] Figure labeling: Module 1 for acquiring sludge treatment process unit set, Module 2 for acquiring multimodal process unit data, Module 3 for acquiring sludge granulation state parameters, and Module 4 for acquiring sludge process unit control parameters. Detailed Implementation

[0011] This application provides a parameter control method and system for improving the efficiency of sludge granulation treatment, which solves the technical problem that the existing technology lacks accurate multimodal monitoring and sludge granulation status assessment, resulting in insufficient efficiency and stability of sludge granulation treatment.

[0012] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0013] It should be noted that the terms "first," "second," etc., in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or modules not explicitly listed or inherent to such processes, methods, products, or devices.

[0014] Example 1, as Figure 1 As shown, a parameter control method for improving the efficiency of sludge granulation treatment is provided, wherein the method includes: Key process flows of the BIOCOS sludge process system are extracted to obtain a set of sludge treatment process units, which consists of a sequence of P-tank units, B-tank units, and SU-tank units.

[0015] In this embodiment, the BIOCOS sludge process system is a modified activated sludge system that achieves efficient nitrogen removal through optimized hydraulic modeling and endogenous denitrification technology. The P tank is a crucial component of the sludge treatment process unit set, with its core function being anaerobic reaction. The B tank is the core reaction unit receiving effluent from the P tank and is a dual-purpose (anoxic / aerobic) tank. The SU tank is a sedimentation and circulation unit, typically two of which operate in an alternating mode.

[0016] Specifically, when extracting the key process flow of the BIOCOS sludge treatment system, the overall operation flow of the system is first comprehensively broken down. Combining the basic logic of functional zoning, material flow, and reaction synergy in activated sludge treatment processes, the core reaction stages and material handling links are identified within the complete chain from influent to effluent. Modified activated sludge systems typically achieve pollutant removal and sludge treatment through the functional division of different tanks. As a type of modified activated sludge system, the BIOCOS process also follows this core framework. Therefore, the key reaction stages—anaerobic reaction, anoxic / aerobic alternating reaction, and sedimentation circulation—are identified sequentially after influent flow, using the material flow direction as a guide.

[0017] Next, based on the aforementioned key reaction stages, we will further analyze the core functions and corresponding tank structures of each stage. In the first key reaction stage after water influent, it is necessary to provide an anaerobic environment for polyphosphate-accumulating bacteria to release phosphorus. This is the core link of the biological phosphorus removal process. In the BIOCOS process, the tank that undertakes this function has independent anaerobic environment control conditions, and the effluent from this tank flows directly to the subsequent reaction tanks. Based on its functional positioning and material flow relationship, this tank is determined to be the first key unit of the sludge treatment process, namely the P-tank unit (anaerobic tank unit).

[0018] Next, the subsequent treatment stage after the effluent from the P tank unit is obtained. This stage requires simultaneous nitrification and denitrification to complete nitrogen removal. This function is usually achieved by connecting an anoxic tank and an aerobic tank in series. However, the BIOCOS process adjusts the aeration time ratio of a single tank, allowing the tank to alternate between anoxic and aerobic states, thus meeting the requirements of both nitrification and denitrification. This dual-state design in one tank is a key feature of the BIOCOS process. By identifying the tank with anoxic / aerobic alternation function and receiving the effluent from the P tank unit, it is determined to be the second key unit of the sludge treatment process, namely the B tank unit (anoxic / aerobic tank unit).

[0019] Subsequently, for the final treatment and sludge recycling of the effluent from Unit B, the sedimentation unit typically only performs solid-liquid separation. However, in the BIOCOS process, the tank in this stage not only needs to complete sludge sedimentation but also needs to achieve sludge return to the aforementioned Unit P and Unit B, sludge-water mixing, and endogenous denitrification. To ensure continuous effluent output, two tanks of this type are usually set up to operate alternately. By identifying the tank that receives the effluent from Unit B, has a combined sedimentation-recycling-endogenous denitrification function, and operates in an alternating mode, it can be determined that this is the third key unit of the sludge treatment process, namely the SU tank unit (sedimentation and recycling tank unit).

[0020] Finally, based on the basic logic of the process unit sequence arrangement, it was verified whether the material flow sequence of each unit conforms to the functional progression relationship of anaerobic phosphorus release-anoxic / aerobic denitrification-sedimentation cycle. It was confirmed that the P tank unit, B tank unit, and SU tank unit form a complete functional chain in the order of influent to effluent, and that the material flow between each unit is smooth and the reaction synergy is strong. Finally, a set of sludge treatment process units composed of the P tank unit, B tank unit, and SU tank unit in the functional progression sequence was obtained.

[0021] By deconstructing the BIOCOS sludge process system, defining its functions, identifying its units, and verifying its sequence, a set of sludge treatment process units consisting of P-tank, B-tank, and SU-tank units was obtained. This provided a precise process unit foundation for deploying monitoring sensors, establishing evaluation channels, and constructing a control strategy library in each unit.

[0022] Multimodal monitoring sensor groups are deployed sequentially on the sludge treatment process unit set to collect data from the sludge treatment multimodal process units.

[0023] Optionally, firstly, a monitoring requirement analysis is performed on each process unit in the sludge treatment process unit set to obtain the monitoring requirement parameters and monitoring parameter boundaries of the sludge process unit. Based on these monitoring requirement parameters and monitoring parameter boundaries, the sensor specification parameters of the sludge process unit are determined. The monitoring key point set of the process unit set is extracted and a coverage analysis is performed to obtain the sensor location parameters of the sludge process unit. Then, based on these sensor specification parameters and sensor location parameters, a multimodal monitoring sensor group is deployed sequentially on the sludge treatment process unit set. The specific execution steps will be explained in detail later.

[0024] Next, within the P-tank unit, the deployed dissolved oxygen sensors, pH sensors, and sludge microbial activity sensors will continue to operate: the dissolved oxygen sensors will capture the dynamic changes of low-concentration dissolved oxygen in real time under anaerobic conditions, the pH sensors will acquire the subtle fluctuations in pH during the anaerobic phosphorus release process, and the sludge microbial activity sensors will reflect the metabolic activity of polyphosphate-accumulating bacteria and other microorganisms during the anaerobic stage; these sensors will convert electrical or digital signals into recordable parameter values ​​according to the set sampling frequency, forming multimodal process data for the P-tank unit.

[0025] The dissolved oxygen sensor, ammonia nitrogen sensor, nitrate nitrogen sensor, oxidation-reduction potential sensor, and sludge concentration sensor in Tank B synchronously collect parameters according to the alternating anoxic / aerobic operation phases: During the aerobic phase, the dissolved oxygen sensor monitors changes in dissolved oxygen concentration in the aeration zone, while the ammonia nitrogen sensor and nitrate nitrogen sensor acquire data on the consumption of ammonia nitrogen and the generation of nitrate nitrogen during the nitrification reaction; After entering the anoxic phase, the oxidation-reduction potential sensor captures the potential transition from aerobic to anoxic conditions, the nitrate nitrogen sensor monitors the degradation of nitrate nitrogen during denitrification, and the sludge concentration sensor continuously acquires sludge concentration data within the tank; each sensor records data with a unified timestamp, generating multimodal process data for Tank B at different reaction stages.

[0026] The sludge concentration sensor, sludge interface monitoring sensor, electromagnetic flow meter, and nitrate nitrogen sensor in the SU tank unit monitor the sludge status, recirculation, and effluent conditions in the sedimentation tank: the sludge concentration sensor acquires the sludge concentration distribution at different depths in the tank in real time; the sludge interface monitoring sensor acquires the height change of the interface during sludge settling, thereby reflecting the sludge settling velocity; the electromagnetic flow meter records the flow rate data of the recirculated sludge; and the nitrate nitrogen sensor measures the residual nitrate nitrogen concentration at the effluent end of the sedimentation tank. These sensors integrate the parameters they collect according to a synchronous acquisition cycle to form multimodal process data for the SU tank unit.

[0027] Ultimately, all the multimodal monitoring sensor groups deployed in the P, B, and SU tanks collect multi-dimensional parameters such as dissolved oxygen, pH, ammonia nitrogen, nitrate nitrogen, sludge concentration, and microbial activity from each tank through the data transmission module. These parameters are then uniformly aggregated into the data acquisition terminal, forming a complete set of multimodal process unit data for sludge treatment. This provides comprehensive and real-time data source support for subsequent sludge granulation status assessment and control strategy optimization.

[0028] A sludge granulation assessment channel is established, and the sludge state is assessed based on the sludge granulation assessment channel for the multimodal process unit data of sludge treatment to obtain sludge granulation state parameters.

[0029] In one embodiment of this application, firstly, historical multimodal process datasets of the multimodal monitoring sensor group are collected, and a standardized processing channel for multimodal process data is constructed based on its characteristic information. Then, a sludge granulation status assessment system is defined, and the historical dataset is used to evaluate and train the fitting to obtain the granulation status assessment channel. Finally, the two channels are connected in series and merged to build a sludge granulation assessment channel. The specific execution steps for building this channel will be described in detail in the following content.

[0030] Subsequently, based on the established sludge granulation assessment channel, the data from the multimodal sludge treatment process units were input into the multimodal process data standardization processing channel within the sludge granulation assessment channel. This multimodal sludge treatment process unit data originated from the multimodal monitoring sensor sets of the P-tank, B-tank, and SU-tank units, covering both numerical and image parameters. The numerical parameters included dissolved oxygen, pH, and sludge microbial activity for the P-tank unit; dissolved oxygen, ammonia nitrogen, nitrate nitrogen, redox potential, and sludge concentration for the B-tank unit; and sludge concentration, sludge interface height, return sludge flow rate, and residual nitrate nitrogen for the SU-tank unit. The image parameters were microscopic images of sludge particles.

[0031] Then, the multimodal process data standardization processing channel adopts an adaptation method for different types of parameters: for numerical parameters, the Z-score standardization method is used to convert the original data into standardized data of a uniform scale by calculating the mean and standard deviation of the parameters; for image parameters, the pixel value normalization method is used to map the pixel values ​​of the sludge particle microscopic image to the [0,1] interval, eliminating the interference of the differences in the dimensions and numerical ranges of different types of parameters on subsequent evaluation, and outputting standardized data in a uniform format.

[0032] Next, the standardized data in a unified format is input into the granulation status assessment channel. This channel consists of multiple parallel branches of a deep feedforward neural network, each corresponding to a specific sludge granulation status assessment index, including sludge volume index, particle size, particle roundness, extracellular polymeric substance content, denitrification rate, and polyphosphate-accumulating organism abundance. Based on the correlation characteristics of each assessment index, the standardized data is precisely allocated to the corresponding branch channel. Nonlinear calculations are then performed on the input standardized data to output predicted values ​​for the corresponding assessment index.

[0033] Subsequently, the predicted values ​​of evaluation indicators output from all deep feedforward neural network branches were summarized and verified. The predicted values ​​were categorized according to the morphological, physicochemical, and functional dimensions of sludge granulation. The morphological dimension included particle size and sphericity; the physicochemical dimension included sludge volume index and extracellular polymeric substance content; and the functional dimension included denitrification rate and polyphosphate-accumulating bacteria abundance. Simultaneously, consistency verification was performed to compare potentially related parameters in the outputs of different branches. For example, sludge concentration data simultaneously affects both sludge volume index and particle stability assessment, ensuring logical consistency and avoiding contradictions among the predicted values. Finally, these values ​​were integrated to form sludge granulation state parameters encompassing multi-dimensional key information.

[0034] By standardizing and eliminating interference from the multimodal process unit data of sludge treatment, the data were input into the parallel deep feedforward neural network branch channels for targeted calculation and summary verification, resulting in comprehensive and accurate sludge granulation state parameters. This provides a clear and reliable state basis for subsequent formulation of sludge granulation treatment efficiency control strategies.

[0035] Based on the sludge treatment process unit set and the sludge granulation treatment target, a sludge process unit control strategy library is constructed. Based on the sludge process unit control strategy library, the sludge granulation state parameters are analyzed and optimized to determine the sludge process unit control parameters. The sludge granulation closed-loop control is then performed using the sludge process unit control parameters.

[0036] Specifically, firstly, the sludge granulation task index set corresponding to the sludge granulation treatment target is extracted. Then, the control logic of each process unit in the sludge treatment process unit set is extracted to obtain the sludge process unit control logic set. Finally, the historical multimodal process dataset is used to perform correlation analysis between the sludge granulation task index set and the sludge process unit control logic set, thereby constructing a sludge process unit control strategy library.

[0037] Next, based on the sludge process unit control strategy library, fuzzy matching optimization is performed on the sludge granulation state parameters to obtain the target sludge process unit control strategy. Then, the target sludge process unit control strategy is used to optimize and analyze the sludge granulation state parameters to determine the sludge process unit control parameters. Finally, sludge granulation closed-loop control is performed using the sludge process unit control parameters.

[0038] Furthermore, the method provided in this application embodiment includes: Monitoring requirements are analyzed sequentially for each process unit in the sludge treatment process unit set to obtain the monitoring requirement parameters and monitoring parameter boundaries of the sludge process units. Based on the monitoring requirement parameters and monitoring parameter boundaries of the sludge process units, the sensor specification parameters of the sludge process units are determined. The monitoring key point set of the sludge treatment process unit set is extracted, and the coverage analysis of the monitoring key point set is performed to obtain the sensor location parameters of the sludge process units. Based on the sensor specification parameters and sensor location parameters of the sludge process units, a multimodal monitoring sensor group is deployed sequentially on the sludge treatment process unit set.

[0039] Specifically, the analysis begins with the core functions and reaction characteristics of each process unit within the BIOCOS sludge treatment system. The P tank, as an anaerobic reaction unit, is primarily responsible for anaerobic phosphorus release from sludge. Therefore, the parameters to be monitored in this unit include dissolved oxygen, pH, and other indicators related to the anaerobic environment and phosphorus release efficiency. The reasonable fluctuation range of these parameters is also determined; for example, dissolved oxygen needs to be maintained at a low level to ensure anaerobic conditions. The B tank, as an anoxic / aerobic alternating reaction unit, undertakes nitrification and denitrification. Key parameters to monitor include dissolved oxygen and sludge concentration, which are related to the switching of reaction stages and denitrification efficiency. Critical values ​​for these parameters at different reaction stages are also defined; for example, dissolved oxygen in the aerobic stage needs to reach a certain concentration to satisfy the activity of nitrifying bacteria. The SU tank, as a sedimentation and circulation unit, is responsible for sludge sedimentation, recirculation, and endogenous denitrification. Parameters related to sedimentation efficiency and recirculation quality, such as sludge concentration and sludge settling velocity, are monitored. The normal operating range of these parameters is also determined. Those skilled in the art can determine and adjust the monitoring parameter boundaries of the sludge treatment units according to the actual project and requirements.

[0040] Next, based on the determined monitoring requirements and boundaries of the sludge process unit, the corresponding sensor specifications for the sludge process unit are further matched. Specifically: To address the monitoring needs of low dissolved oxygen in the P-tank, a dissolved oxygen sensor with a measurement range covering the low concentration range and meeting the required accuracy was selected to ensure accurate capture of subtle changes in dissolved oxygen under anaerobic conditions. pH value is a key indicator reflecting the anaerobic phosphorus release environment; a pH sensor with high measurement accuracy and stable operation under anaerobic acidic or neutral conditions was selected to accurately capture subtle pH fluctuations during the anaerobic reaction. Simultaneously, to monitor the metabolic activity of sludge microorganisms during the anaerobic phosphorus release stage, a sludge microbial activity sensor adapted to the anaerobic, low-disturbance environment of the P-tank was chosen. Through these matching processes, the sensor type, measurement range, and environmental adaptability specifications for monitoring indicators such as dissolved oxygen, pH value, and sludge microbial activity in the P-tank were determined.

[0041] To monitor the differences in dissolved oxygen at different reaction stages in Tank B, a dissolved oxygen sensor with a fast response speed and adaptability to switching between aeration and anoxic environments was selected to avoid data deviation due to sensor response lag. Ammonia nitrogen and nitrate nitrogen concentrations are core indicators reflecting nitrification and denitrification efficiency; therefore, ammonia nitrogen and nitrate nitrogen sensors with fast response speeds and measurement ranges covering concentration changes during the denitrification process were selected. Oxidation-reduction potential (ORP) can help determine the timing of the switch between anoxic and aerobic stages; therefore, an ORP sensor capable of stable operation under alternating aeration and stirring environments was chosen. The sludge concentration sensor needed to adapt to the sludge mixing state caused by aeration and stirring in Tank B and possess anti-interference capabilities. Based on this, the sensor specifications for monitoring indicators such as dissolved oxygen, ammonia nitrogen, nitrate nitrogen, ORP, and sludge concentration in Tank B were determined.

[0042] For monitoring sludge concentration in the SU tank, a sludge concentration sensor resistant to sludge adhesion and suitable for the complex environment of the sedimentation tank was selected to ensure the stability of long-term monitoring data. Through this precise matching of parameters and specifications, key specifications such as sensor type, measurement range, and accuracy for each monitoring indicator can be determined. Sludge settling velocity is obtained using a sludge interface monitoring sensor capable of accurately capturing changes in the sludge interface height in the sedimentation tank, and this sensor must be adaptable to the alternating flow patterns of the SU tank. For monitoring the return sludge flow rate, an electromagnetic flowmeter capable of stable operation in the sludge return pipeline is selected. The residual nitrate nitrogen after endogenous denitrification in the SU tank is an important indicator for evaluating the denitrification effect, requiring a sensor suitable for measuring low-concentration nitrate nitrogen at the sedimentation tank effluent. Therefore, the sensor specifications for monitoring indicators such as sludge concentration, sludge interface (correlated with settling velocity), return sludge flow rate, and residual nitrate nitrogen in the SU tank are determined.

[0043] Then, when extracting the key monitoring points for each sludge treatment process unit, focus on the locations within each process unit that significantly affect the reaction process and sludge granulation effect. Key monitoring points for the P tank should include the raw water inlet and the mixing point between the returned sludge and raw water, as these locations directly affect the initial conditions and reaction uniformity of anaerobic phosphorus release. Key monitoring points for the B tank need to cover the aeration zone, the boundary between the anoxic zone and the effluent end, as these locations reflect parameter changes at different reaction stages, helping to determine the timing of nitrification and denitrification switching. Key monitoring points for the SU tank should include the bottom of the sedimentation tank, the sludge return outlet, and the effluent area; bottom parameters reflect the sludge settling effect, while return outlet parameters relate to the quality of the returned sludge.

[0044] Next, after extracting these key monitoring points, a coverage analysis is conducted to check whether the key points can fully capture the spatial distribution and dynamic changes of parameters within the unit, avoiding monitoring blind spots. For example, if only the top area of ​​the SU tank is monitored, situations where the sludge concentration at the bottom is too high or too low may be missed. By adjusting the distribution of key monitoring points, it is ensured that the monitoring range can cover the core reaction area of ​​the entire unit, thereby obtaining the sensor location parameters of the sludge process unit.

[0045] Finally, based on the determined sensor specifications and location parameters for the sludge process units, multimodal monitoring sensor groups were deployed sequentially according to the process unit sequence. In tank P, dissolved oxygen and pH sensors of the correct specifications were installed at key locations such as the mixing area; in tank B, dissolved oxygen and sludge concentration sensors of different specifications were installed at key points such as the aerobic zone, anoxic zone, and boundary; in tank SU, sludge concentration and sludge settling velocity monitoring sensors were installed at the bottom and return outlet. During deployment, it was ensured that the sensors were compatible with the operating environment of the process units, such as avoiding severe impact from air bubbles on sensors in the aeration area of ​​tank B, and preventing sludge clogging of sensors at the bottom of tank SU. This completed the deployment of the multimodal monitoring sensor groups across the entire sludge treatment process unit set.

[0046] By sequentially conducting process unit monitoring requirement analysis, determining sensor specifications and parameters, extracting key monitoring points and performing coverage analysis, and deploying a multimodal monitoring sensor group, accurate and comprehensive monitoring of key reaction parameters of each unit in the P, B, and SU tanks of the BIOCOS process was achieved. This provides reliable and continuous multimodal process data support for subsequent sludge granulation status assessment and control strategy optimization.

[0047] Furthermore, the method provided in this application embodiment includes: Historical multimodal process datasets of the multimodal monitoring sensor group are collected; based on the characteristic information of the historical multimodal process datasets, standardized steps are analyzed to construct a multimodal process data standardization processing channel; a sludge granulation state assessment system is defined, and the sludge granulation state assessment system is used to evaluate, train, and fit the historical multimodal process datasets to obtain a granulation state assessment channel; the multimodal process data standardization processing channel and the granulation state assessment channel are cascaded and merged to construct the sludge granulation assessment channel.

[0048] Optionally, firstly, historical multimodal process datasets are collected from the multimodal monitoring sensor group. This group includes dissolved oxygen sensors, pH sensors, sludge concentration sensors, ammonia nitrogen sensors, nitrate nitrogen sensors, and sludge particle image acquisition devices, such as online microscopic imaging instruments. During the operation of the BIOCOS sludge process system, these sensors and devices continuously monitor parameters such as dissolved oxygen, pH, and sludge microbial activity in the P tank unit; dissolved oxygen, ammonia nitrogen, nitrate nitrogen, redox potential, and sludge concentration in the B tank unit; and sludge concentration, sludge interface height, and return sludge flow rate in the SU tank unit. From the historical database of this sensor group, different operating cycles are retrieved according to time series, such as the sludge granulation start-up period, stabilization period, and optimization adjustment period; historical data under different influent water quality conditions are also retrieved, such as influent COD concentration fluctuating between 300 mg / L and 800 mg / L, and ammonia nitrogen concentration fluctuating between 30 mg / L and 60 mg / L, forming a historical multimodal process dataset containing both numerical and image parameters.

[0049] Next, based on the characteristic information of the historical multimodal process dataset, a standardization step analysis is performed to construct a multimodal process data standardization processing channel: First, the numerical parameters in the historical multimodal process dataset are analyzed for characteristics. For example, parameters such as dissolved oxygen, ammonia nitrogen, and nitrate nitrogen are continuous values, and their parameter ranges differ between different process units. For instance, dissolved oxygen in the P tank unit is typically 0-0.5 mg / L, while in the aerobic section of the B tank unit it is 2-4 mg / L. pH values ​​are continuous values ​​that fluctuate within a certain range. For these numerical parameters, the Z-score standardization method is used, and its formula is: Where x is the original data, The mean of the data. This represents the standard deviation of the data. For example, for dissolved oxygen data in the aerobic section of Pool B, assuming the mean is 3 mg / L and the standard deviation is 1 mg / L, if the dissolved oxygen at a certain moment is 4 mg / L, the Z-score standardized value is (4-3) / 1=1.

[0050] For image-based parameters such as sludge particle images, an image standardization method is employed. First, the image is converted to grayscale, and then pixel values ​​are normalized, mapping them to the range of 0-1. For example, if the pixel value range in the original microscopic image is 0-255, after normalization, the value of a certain pixel is 128 / 255 = 0.502. By selecting appropriate standardization methods for different types of parameters, the historical multimodal process dataset is standardized, thus constructing a multimodal process data standardization processing channel. This channel can convert the originally collected multi-dimensional and multi-type process data into data in a unified standard format, facilitating the training and calculation of subsequent sludge particle assessment models.

[0051] Next, the indicators of the sludge granulation status assessment system are extracted to obtain the sludge granulation status assessment indicator set. Each assessment indicator in the assessment indicator set is sequentially associated and mapped with the historical multimodal process dataset to obtain the sludge status indicator association feature dataset. Based on the association feature dataset, assessment training and fitting are performed to generate the sludge status indicator assessment branch channel set. Then, the assessment branch channel set is combined in parallel to obtain the granulation status assessment channel. The specific steps will be explained in detail in the following content.

[0052] Then, during the cascading and merging of the two channels mentioned above, the functional positioning and data flow logic of the multimodal process data standardization processing channel and the granular state assessment channel are first clarified. The multimodal process data standardization processing channel receives the raw data collected by the multimodal monitoring sensor group. This data covers multi-dimensional information such as dissolved oxygen and pH value of the P tank unit, ammonia nitrogen, nitrate nitrogen, and redox potential of the B tank unit, and sludge concentration, sludge interface height, and sludge particle microscopic images of the SU tank unit. This channel transforms raw data of different dimensions and types into a unified standard format through Z-score standardization and image normalization, such as mapping numerical parameters to the [-1,1] interval and normalizing image pixel values ​​to the [0,1] interval, eliminating the interference of data scale differences on subsequent assessments.

[0053] Subsequently, the output data from the multimodal process data standardization channel is used as the input data for the granulation status assessment channel and then concatenated. The granulation status assessment channel is constructed based on a deep feedforward neural network and includes independent branches for assessment indicators such as sludge volume index, particle size, particle sphericity, and extracellular polymer content. Standardized data in a unified format is precisely allocated according to the input requirements of each branch. For example, standardized sludge concentration and sludge interface height data for the SU tank unit are allocated to the sludge volume index assessment branch, while standardized dissolved oxygen and ammonia nitrogen degradation rate data for the B tank unit are allocated to the particle size assessment branch. Each branch calculates the input data using model parameters fitted during pre-training, such as the number of hidden layer neurons, activation function type, and optimizer parameters, and outputs predicted values ​​for the corresponding assessment indicators. The outputs of all branch channels are then summarized to form a complete sludge granulation status assessment report.

[0054] By constructing a standardized processing channel for multimodal process data and a training and fitting channel to obtain a granulation state assessment channel, and then merging the two channels in series, a sludge granulation assessment channel is built, providing a complete data processing and assessment system to support the subsequent accurate assessment of sludge granulation state.

[0055] Furthermore, the method provided in this application embodiment includes: The sludge granulation state assessment system is subjected to index extraction to obtain a sludge granulation state assessment index set; each assessment index in the sludge granulation state assessment index set is sequentially associated and mapped with the historical multimodal process dataset to obtain a sludge state index association feature dataset; assessment training and fitting are performed based on the sludge state index association feature dataset to generate a sludge state index assessment branch channel set; the sludge state index assessment branch channel set is combined in parallel to obtain the granulation state assessment channel.

[0056] Specifically, when defining the sludge granulation status assessment system, we focus on the core processes such as the formation, development, and stability of sludge particles. We combine the correlation between sludge granulation and particle morphology characteristics, physicochemical properties, microbial community characteristics, and process operating parameter responses in biological treatment processes to build a sludge granulation status assessment system that covers multiple aspects, thus clarifying the overall direction for subsequent indicator extraction.

[0057] Next, the evaluation system was analyzed by extracting indicators: for particle morphology, particle size was extracted and measured using online microscopic imaging combined with image segmentation algorithms to statistically determine the proportion of particles in different size ranges; particle roundness was determined using the formula... The calculations reflect the degree to which particles are nearly spherical; the fractal dimension of particles (which reflects the complexity of particle structure), and other indicators.

[0058] For the physicochemical properties dimension, combined with the evaluation method of sludge settling and dewatering performance, the sludge volume index (SVI), which is the volume of a unit mass of sludge after standing for 30 minutes; sludge density, which reflects the compactness of the particles; and extracellular polymeric substances (EPS), with a focus on protein and polysaccharide content, which play a key role in particle stability.

[0059] For the microbial community characteristics dimension, molecular biological detection methods are used to extract the abundance of functional bacteria, such as the relative abundance of polyphosphate-accumulating bacteria and nitrifying bacteria, and to determine them by fluorescence quantitative PCR or high-throughput sequencing; microbial diversity indices, such as the Shannon index, reflect the level of community diversity and other indicators.

[0060] For the process operation parameter response dimension, based on the correlation analysis of the sludge treatment process unit operation data, the tolerance of granulated sludge to influent load is extracted, which is reflected by the stability of the system effluent water quality under different influent COD (chemical oxygen demand) loads; indicators such as the denitrification rate of granulated sludge and the amount of nitrate nitrogen degradation per unit time during the anoxic stage of the B tank unit are also included. Through detailed decomposition and extraction of each of the above dimensions, a comprehensive set of sludge granulation status assessment indicators covering key aspects of sludge granulation status is formed.

[0061] Next, sludge features are extracted from the historical multimodal process dataset to obtain a static feature set and a dynamic feature set of the sludge process. Based on these two feature sets, a cascade analysis is performed to generate a sludge process feature recursion tree. A key analysis is performed on each evaluation index in the sludge granulation status evaluation index set to obtain a key coefficient set of sludge evaluation indicators. Based on this coefficient set, a hierarchical set of associated features of sludge evaluation indicators is determined. Then, based on this hierarchical set, it is sequentially associated and mapped with the sludge process feature recursion tree to obtain a sludge status index associated feature dataset. This step will be explained in detail in the following sections.

[0062] Then, for each evaluation index in the sludge granulation status evaluation index set, a corresponding deep feedforward neural network model is constructed. The deep feedforward neural network has the ability to handle multi-dimensional nonlinear feature correlations, can accurately fit the complex mapping relationship between process features and single evaluation indexes, and is suitable for the multi-parameter coupling scenario of P tank, B tank and SU tank units in the BIOCOS process.

[0063] For each evaluation indicator, the associated feature dataset of sludge state indicators is as follows: for example, the associated feature dataset of sludge volume index includes sludge concentration, sludge settling velocity, and sedimentation area of ​​SU tank; the associated feature dataset of particle size includes dissolved oxygen concentration, ammonia nitrogen degradation rate, and effective volume of B tank. First, the dataset is divided into 70% training set, 15% validation set, and 15% test set to ensure that the data distribution conforms to the actual operating conditions of the BIOCOS process, such as influent load fluctuations and aeration time adjustments.

[0064] Subsequently, a deep feedforward neural network structure corresponding to a single index was constructed: taking the evaluation branch of the sludge volume index as an example, the number of nodes in the input layer was set to 3, corresponding to three related features: sludge concentration in the SU tank, sludge settling velocity, and sedimentation area in the SU tank; two hidden layers were set, with 16 neurons in the first layer and 8 neurons in the second layer, and the ReLU function was used as the activation function to enhance the model's ability to fit nonlinear relationships; the output layer had 1 neuron, and the output was the predicted value of the sludge volume index. During training, the mean squared error (MSE) was used as the loss function to measure the deviation between the model's predicted value and the actual sludge volume index value in the dataset; the Adam optimizer was selected, with the initial learning rate set to 0.001 and the number of iterations set to 500 rounds, and an early stopping strategy was introduced, that is, training was stopped when the validation set loss did not decrease for 10 consecutive rounds to avoid model overfitting.

[0065] Subsequently, for the associated feature datasets of other evaluation indicators, deep feedforward neural network branch channels were constructed according to the above logic. After each branch channel was trained, the model performance was verified through a test set. The coefficient of determination R² between the predicted and actual values ​​of each evaluation indicator was required to be no less than 0.9, ensuring that the evaluation accuracy of the branch channels met the actual requirements of those skilled in the art for accurate diagnosis of granulation status. For example, the particle size prediction error was controlled within ±0.05mm, and the sludge volume index prediction error was controlled within ±5mL / g.

[0066] After training and validating all sludge condition assessment branch channels, the branch channels are combined in parallel to obtain a granular condition assessment channel. The core of the parallel combination is to retain the independent computational logic of each deep feedforward neural network branch channel, while constructing a unified input interface and output aggregation module: the unified input interface receives standardized data processed by the multimodal process data standardization processing channel, including multi-dimensional standardized process data such as dissolved oxygen, pH, sludge concentration, ammonia nitrogen, and nitrate nitrogen from the P tank, B tank, and SU tank units, and distributes the data to the corresponding branches according to the input requirements of each branch channel.

[0067] The output summary module collects the output results of all branch channels, such as the predicted values ​​of assessment indicators like sludge volume index, particle size, particle roundness, extracellular polymeric substances, denitrification rate, and polyphosphate abundance, to form a complete sludge granulation status assessment report, thereby comprehensively covering the morphological, physicochemical, and functional characteristics of sludge granulation.

[0068] By constructing a deep feedforward neural network branch channel for each sludge granulation assessment index and conducting targeted training, and then combining the branch channels in parallel, a granulation status assessment channel based on a neural network model was formed. This enabled multi-dimensional and high-precision synchronous assessment of the sludge granulation status of the BIOCOS sludge process system, providing a reliable status diagnosis basis for subsequent process control strategy optimization.

[0069] Furthermore, the method provided in this application embodiment includes: Sludge features are extracted from the historical multimodal process dataset to obtain a static feature set and a dynamic feature set of the sludge process. A cascade analysis is performed based on these two sets to generate a sludge process feature recursion tree. A criticality analysis is conducted on each evaluation index in the sludge granulation state evaluation index set to obtain a set of critical coefficients for the sludge evaluation index. Based on this set of critical coefficients, a hierarchical set of associated features for the sludge evaluation index is determined. Finally, the hierarchical set of associated features for the sludge evaluation index is sequentially mapped to the sludge process feature recursion tree to obtain the sludge state index associated feature dataset.

[0070] Specifically, when extracting sludge features from historical multimodal process datasets, these datasets encompass multidimensional parameters of the P-tank, B-tank, and SU-tank units across different operating cycles, including both numerical and graphical parameters. First, the static sludge process feature set is obtained by extracting parameters inherent to the process units that do not dynamically change during operation, such as the effective volume of the P-tank, B-tank, and SU-tank units, the tank area division parameters corresponding to sensor installation locations, and the internal partition wall structure parameters of the tanks.

[0071] The dynamic feature set of sludge process is obtained by extracting parameters that change in real time during operation, such as the dissolved oxygen concentration fluctuation and pH value change curve of the P tank unit, the dissolved oxygen switching data between the aerobic and anoxic sections of the B tank unit, the degradation rate of ammonia nitrogen and nitrate nitrogen, the real-time monitoring value of sludge concentration, the change of sludge interface height, the dynamic data of sludge return flow rate of the SU tank unit, and the real-time statistical value of particle size and the trend of particle roundness change in the sludge particle microscopic image, etc., which ultimately form two feature sets corresponding to static and dynamic characteristics respectively.

[0072] Next, a cascade analysis was conducted based on the static and dynamic feature sets of the sludge process to establish the correlation logic between these two types of features: taking the P tank unit as an example, the effective volume of the P tank unit (static feature) was correlated with the fluctuation of dissolved oxygen concentration and the change curve of pH value in the tank (dynamic feature) to analyze the synergistic effect of dissolved oxygen and pH value on the anaerobic phosphorus release effect under different volumes; for the B tank unit, the layout parameters of the aeration device in the B tank unit (static feature) were correlated with the aeration time ratio and the ammonia nitrogen degradation rate (dynamic feature) to explore the relationship between the device layout and the aeration efficiency and nitrification reaction; for the SU tank, the sedimentation area of ​​the SU tank (static feature) was correlated with the sludge settling velocity and the change of sludge interface height (dynamic feature) to clarify the influence of the sedimentation space on the sludge solid-liquid separation effect.

[0073] During the cascade analysis, a recursive partitioning method was used to split the feature relationships according to the process unit level. First, the P-B-SU tank was taken as the first-level branch, and then the static feature-dynamic feature second-level sub-branch was subdivided under each branch. Then, the specific parameter association sub-items were further decomposed under the second-level sub-branch, and finally a sludge process feature recursive tree with the process unit as the core and the static and dynamic features deeply coupled was generated.

[0074] Then, a criticality analysis was conducted on each evaluation indicator in the sludge granulation status assessment indicator set, which includes indicators such as sludge volume index, particle size, particle roundness, extracellular polymeric substance content, denitrification rate, and polyphosphate-accumulating bacteria abundance. The Pearson correlation coefficient method was used to calculate the correlation between each evaluation indicator and the core objectives of sludge granulation, such as granulation rate, stable operation cycle of granules, and pollutant removal efficiency. Simultaneously, grey relational analysis was used to quantify the influence of each indicator on the granulation effect: for example, the correlation coefficient between sludge volume index and granulation rate was calculated; if the absolute value of the correlation coefficient is close to 1, it indicates that the indicator is extremely critical to the granulation status assessment. The grey relational degree between particle size and pollutant removal efficiency was analyzed; a high correlation degree indicates that the indicator is crucial to assessing granule function. Through the above methods, a criticality coefficient set of sludge evaluation indicators reflecting the importance of each indicator was finally obtained.

[0075] Subsequently, when determining the hierarchical set of associated features for sludge assessment indicators based on the set of critical coefficients, the assessment indicators are sorted from highest to lowest critical coefficient. Indicators with higher critical coefficients are classified into higher levels, while those with lower critical coefficients are classified into lower and middle levels. For example, if the sludge volume index has the highest critical coefficient, its associated features include the dynamic characteristics of sludge concentration and sludge settling velocity in the SU tank unit, as well as the static characteristics of the sedimentation area in the SU tank unit; these features are classified into the first level. The particle size index has the next highest critical coefficient, and its associated features include the dynamic characteristics of dissolved oxygen concentration and ammonia nitrogen degradation rate in the B tank unit, as well as the static characteristics of the effective volume in the B tank unit; these are classified into the second level. If the polyphosphate accumulation abundance index has a low critical coefficient, its associated features include the dynamic characteristics of pH value and the static characteristics of anaerobic retention time in the P tank unit; these are classified into the third level. In this way, a corresponding feature level is matched to each assessment indicator, forming a clearly structured hierarchical set of associated features for sludge assessment indicators.

[0076] Finally, for each evaluation indicator's associated feature level, the corresponding branch node is located in the feature recursion tree. For example, the associated feature level of the sludge volume index corresponds to the second-level sub-branches of dynamic features (sludge concentration, sludge settling velocity) and static features (sedimentation area) under the first-level branch of the SU tank unit in the recursion tree. Specific feature data under these sub-branches are extracted and associated with the sludge volume index. Similarly, the associated feature level of particle size corresponds to the second-level sub-branches of dynamic features (dissolved oxygen concentration, ammonia nitrogen degradation rate) and static features (effective volume) under the first-level branch of the B tank unit in the recursion tree. Feature data is also extracted and associated with the particle size index. Following the above logic, the mapping of all sludge evaluation indicators to recursion tree features is completed. The feature data corresponding to each indicator is integrated with the evaluation data of the indicator itself, ultimately resulting in a sludge state indicator associated feature dataset containing evaluation indicators, associated features, and feature data.

[0077] By extracting features from historical multimodal process data, constructing recursive trees through cascade analysis, quantifying the key aspects of evaluation indicators, dividing feature levels, and mapping relationships, a sludge state indicator association feature dataset was formed that can accurately associate sludge granulation evaluation indicators with process characteristics. This provides structured and targeted data support for the subsequent training and fitting of the granulation state evaluation channel.

[0078] Furthermore, the method provided in this application embodiment includes: Extract the sludge granulation task index set to obtain the sludge granulation treatment target; extract the control logic of each process unit in the sludge treatment process unit set to obtain the sludge process unit control logic set; use the historical multimodal process dataset to perform correlation analysis on the sludge granulation task index set and the sludge process unit control logic set to construct a sludge process unit control strategy library.

[0079] Specifically, firstly, when extracting the set of sludge granulation task indicators to obtain the target of sludge granulation treatment, it is necessary to first clarify the core objectives of sludge granulation treatment, which typically revolve around improving granulation efficiency, ensuring long-term stable operation of granules, and simultaneously optimizing pollutant removal effects. Referring to industry technical specifications familiar to those skilled in the art and the practical parameters of the BIOCOS process in sludge granulation modification, and combining the correlation between key characteristics and treatment effects during the sludge granulation process, quantifiable task indicators are determined.

[0080] For example, regarding granulation efficiency, a granulation rate index is extracted, with a target value set at ≥80%, i.e., the proportion of particles with a diameter ≥0.3mm to the total sludge volume. Regarding particle stability, particle size distribution is extracted, with the effective particle size range controlled between 0.3-1.2mm, and the sludge volume index controlled between 30-50mL / g to ensure good settling performance. The protein-to-polysaccharide ratio of extracellular polymeric substances is ≥2.5 to enhance particle structural stability. Regarding pollutant removal function, indicators such as denitrification rate and the ratio of anaerobic phosphorus release to aerobic phosphorus uptake in polyphosphate accumulation efficiency are extracted, with a target value ≥0.8. Each indicator corresponds to a key dimension of the sludge granulation treatment target. By integrating these quantifiable and monitorable indicators, a sludge granulation task indicator set is ultimately formed, providing clear target guidance for subsequent process control.

[0081] Next, when extracting the control logic for each process unit in the sludge treatment process unit set to obtain the sludge process unit control logic set, the process is developed by combining the functional positioning and microbial metabolic characteristics of each process unit, and relying on the correlation between parameter adjustment and treatment effect in historical multimodal process data. The sludge treatment process unit set includes P tank, B tank, and SU tank units: For the P-tank unit, its core function is to provide an anaerobic phosphorus release environment for polyphosphate-accumulating bacteria. Based on the principle of anaerobic metabolism and historical data on the impact of dissolved oxygen and pH value of the P-tank unit on the subsequent granulation effect, control logic is extracted. For example, the dissolved oxygen of the P-tank unit needs to be maintained at 0.1-0.3 mg / L. When the monitored value exceeds 0.3 mg / L, the influent flow rate is reduced by 10%-15% or the stirring intensity is increased from 30 r / min to 40 r / min until the dissolved oxygen returns to the target range. At the same time, the pH value is controlled at 7.0-7.5. When the pH < 7.0, it is adjusted by adding sodium bicarbonate solution, with the addition amount calculated at 0.5 g / L for every 0.1 pH unit decrease.

[0082] For tank B, nitrification and denitrification need to be alternated to ensure nitrogen removal efficiency and promote particle formation. Based on the physiological characteristics of nitrifying bacteria (aerobic) and denitrifying bacteria (anoxic), control logic is implemented. For example, dissolved oxygen is controlled at 2-4 mg / L during the aerobic stage, and the anoxic stage is switched when the ammonia nitrogen concentration is <5 mg / L. During the anoxic stage, the oxidation-reduction potential is maintained at -100 to -50 mV for 1.5-2 hours before switching back to the aerobic stage. If the nitrate nitrogen concentration is <2 mg / L, the anoxic stage is ended early.

[0083] For the SU tank unit, the main responsibility is sludge particle settling and recirculation. Based on particle settling kinetics, control logic is extracted. For example, the sludge interface height in the SU tank unit needs to be controlled at 60%-70% of the tank height. When the interface height exceeds 70%, the recirculation ratio is increased from 300% to 350%. When the interface height is below 60%, the daily sludge discharge is reduced from 5% of the tank volume to 3%, while monitoring the recirculated sludge concentration. If the concentration is <6g / L, the influent flow rate is reduced to increase the sludge concentration in the tank. By decomposing, verifying, and sorting out the control logic of each of the three process units, a sludge process unit control logic set covering the core control requirements of each unit is finally formed.

[0084] Finally, the sludge granulation task index set, the sludge process unit control logic set, and the historical multimodal process dataset are associated and separated to obtain the sludge process unit control logic dataset and the sludge granulation task index dataset. The control data of each process unit in the sludge process unit control logic dataset are associated and analyzed with the sludge granulation task index dataset to obtain the sludge process unit strategy controller set. Then, the controller set is optimized and fused with multiple objectives to construct the sludge process unit control strategy library. This step will be explained in detail in the following sections.

[0085] By referencing industry standards and process practices, a set of quantifiable sludge granulation task indicators was extracted. Combined with the microbial metabolic principles of each process unit and historical operating data, targeted control logic was extracted. This laid a clear and feasible foundation for subsequent correlation analysis using historical multimodal process datasets and the construction of a sludge process unit control strategy library.

[0086] Furthermore, the method provided in this application embodiment includes: The sludge granulation task index set and the sludge process unit control logic set are associated and separated with the historical multimodal process dataset to obtain the sludge process unit control logic dataset and the sludge granulation task index dataset. The control data of each process unit in the sludge process unit control logic dataset are associated with the sludge granulation task index dataset to obtain the sludge process unit strategy controller set. The sludge process unit strategy controller set is then optimized and fused with multiple objectives to construct a sludge process unit control strategy library.

[0087] In one embodiment, firstly, the core content and corresponding relationships of the three types of data are clarified. The historical multimodal process dataset covers multi-dimensional parameter records of the P-tank, B-tank, and SU-tank units in different operating cycles, including numerical and graphical parameters, and each data point is accompanied by a timestamp and the identifier of its respective process unit. The sludge granulation task indicator set includes quantifiable target indicators such as granulation rate, particle size distribution, sludge volume index, extracellular polymeric substance content, denitrification rate, and polyphosphate efficiency; the sludge process unit control logic set consists of the control rules for the P-tank, B-tank, and SU-tank units, such as the dissolved oxygen control range for the P-tank, the anoxic / aerobic switching conditions for the B-tank, and the sludge return ratio adjustment logic for the SU-tank.

[0088] During the execution of the associated diversion process, timestamps and process unit identifiers are used as the matching criteria: Parameter data directly related to the control rules of each process unit in the historical multimodal process dataset are associated with the corresponding unit rules in the sludge process unit control logic set. For example, dissolved oxygen adjustment records and pH adjustment data of the P tank unit in historical data are associated with the P tank unit control logic; aeration duration and oxidation-reduction potential change data of the B tank unit are associated with the B tank unit control logic; and return pump frequency and sludge discharge data of the SU tank unit are associated with the SU tank unit control logic, thus forming a sludge process unit control logic dataset. Simultaneously, parameter data reflecting the actual achievement of granulation task indicators in the historical multimodal process dataset are associated with the corresponding indicators in the sludge granulation task indicator set. For example, particle size statistics in historical data are associated with particle size distribution indicators; measured sludge volume index values ​​are associated with sludge volume index indicators; and calculated denitrification rates are associated with denitrification rate indicators, thus forming a sludge granulation task indicator dataset. This ensures that the two types of datasets correspond to the control logic execution process and the achievement of task indicators, respectively.

[0089] Subsequently, correlation analysis was performed on the control data of each process unit in the sludge process unit control logic dataset and the sludge granulation task indicator dataset. Analysis was conducted separately for each process unit, using data association algorithms to uncover the intrinsic relationship between control data and the achievement of task indicators. Taking the P-tank unit as an example, dissolved oxygen control data (such as the duration and adjustment frequency of different dissolved oxygen ranges) and pH control data (such as pH adjustment amplitude and the percentage of time maintained within the target range) were extracted from the sludge process unit control logic dataset. These were then matched with indicators such as polyphosphate efficiency and granulation rate in the sludge granulation task indicator dataset. The Pearson correlation coefficient method was used to calculate the correlation between control parameters and indicators. If the analysis found that when dissolved oxygen in the P-tank was maintained at 0.1-0.3 mg / L and pH was maintained at 7.0-7.5, the probability of polyphosphate efficiency reaching ≥0.8 and granulation rate reaching ≥80% was significantly higher than in other ranges, then the correspondence between this set of control parameters and indicator achievement results was taken as the core control correlation law for the P-tank.

[0090] Next, the correlation analysis is performed on the B tank unit and the SU tank unit in the same manner as the P tank unit. Finally, the control correlation rules of each process unit are transformed into executable control logic modules, such as the dissolved oxygen-pH coordinated control module for the P tank unit, the anoxic / aerobic switching control module for the B tank unit, and the reflux ratio-interface height control module for the SU tank unit, which are integrated to form a sludge process unit strategy controller set.

[0091] Finally, inter-unit influence analysis and control coupling are performed on the sludge process unit strategy controller set to generate a sludge process unit strategy coupled controller. Based on the sludge granulation task index set, a sludge granulation treatment objective function is constructed. Then, based on this treatment objective function, multi-objective solution and strategy optimization fusion are performed on the sludge process unit strategy coupled controller to construct a sludge process unit control strategy library. This step will be explained in detail in the following sections.

[0092] By associating and diverting data based on timestamps and process unit identifiers, and then conducting correlation analysis between control data and task indicator data for each process unit, a unit-based, strongly correlated basic data and controller module support is provided for the subsequent multi-objective optimization and integration to build a sludge process unit control strategy library.

[0093] Furthermore, the method provided in this application embodiment includes: Inter-unit influence analysis and control coupling are performed on the sludge process unit strategy controller set to generate a sludge process unit strategy coupled controller; based on the sludge granulation task index set, a sludge granulation treatment objective function is constructed; based on the sludge granulation treatment objective function, multi-objective solution and strategy optimization fusion are performed on the sludge process unit strategy coupled controller to construct the sludge process unit control strategy library.

[0094] Optionally, based on the process flow correlation, the interaction relationships between the controllers of each process unit are first identified. Starting from the process logic, the impact of parameter adjustments of the upstream unit controllers on the operating status of subsequent units is analyzed, as well as the synergistic or restrictive relationships of parameter changes between parallel units, clarifying the correlation paths between the output parameters of each controller and the input conditions of other units. On this basis, control coupling is carried out, integrating the control logic of the dispersed process unit strategy controllers. For control parameters that conflict or are related, parameter coordination adjustment rules are established based on process priority and treatment target requirements to eliminate control contradictions between different controllers, forming a sludge process unit strategy coupling controller that can coordinate the operation of each unit and achieve parameter linkage adjustment, ensuring that the control actions of each unit cooperate with each other rather than operate independently.

[0095] Next, based on the sludge granulation task indicator set, an objective function for sludge granulation treatment is constructed, transforming the various quantitative indicators in the task indicator set into calculable mathematical expressions. For each indicator, its optimization direction is clarified, distinguishing between maximization and minimization objectives. Then, considering the importance of each indicator in sludge granulation treatment, the weight coefficients of each indicator are determined using an expert scoring method. Simultaneously, constraints on the objective function are set based on the actual operational limitations of the sludge treatment process, covering the parameter operating range of each process unit, energy consumption, and maximum reagent dosage. The mathematical expressions of each indicator are integrated according to their weight coefficients to form a multi-objective function with the core objective of comprehensively improving the sludge granulation effect. This function should intuitively reflect the correlation between the achievement of each task indicator and the overall treatment effect.

[0096] Then, based on the objective function of sludge granulation treatment, a multi-objective solution and strategy optimization fusion were performed on the strategy-coupled controller of the sludge process unit. A multi-objective optimization algorithm was used for the solution calculation. The algorithm iteratively searched for various combinations of control parameters in the coupled controller, calculated the comprehensive score of the objective function under different parameter combinations, and selected multiple sets of control parameter combinations that satisfied the constraints and maximized the objective function score.

[0097] Finally, the selected optimal parameter combinations are transformed into strategies, converting each combination into specific, executable process unit control strategies. This clarifies the parameter adjustment methods and triggering conditions for each unit under different operating conditions. These control strategies are then integrated, removing duplicate or conflicting strategies and adding transition rules between strategies. This results in a sludge process unit control strategy library that covers different operating scenarios and can be flexibly invoked. This ensures that the strategies in the library not only meet multi-objective optimization requirements but also adapt to the dynamic adjustment needs of actual processes.

[0098] By conducting inter-unit impact analysis and control coupling, constructing multi-objective functions, and using optimization algorithms to solve and integrate them, a sludge process unit control strategy library that coordinates all process units and meets multi-objective requirements has been formed, providing systematic and optimized strategy support for parameter regulation in subsequent sludge granulation treatment.

[0099] Furthermore, the method provided in this application embodiment includes: Based on the sludge process unit control strategy library, fuzzy matching optimization is performed on the sludge granulation state parameters to obtain the target sludge process unit control strategy; the target sludge process unit control strategy is then used to perform control optimization analysis on the sludge granulation state parameters to determine the sludge process unit control parameters.

[0100] In one embodiment, the core features of the sludge granulation state parameters are first extracted. These features must be consistent with the preset key feature dimensions of each control strategy in the sludge process unit control strategy library to ensure the uniformity of the matching dimensions. Then, proximity calculation is used as a fuzzy matching method. By calculating the proximity between the current sludge granulation state parameter features and the key features of each control strategy in the control strategy library, the similarity between the two is quantified. During the calculation process, similarity comparison is directly performed based on the preset feature dimensions, and the control strategy with the highest proximity is selected and determined as the target sludge process unit control strategy, ensuring that the matching process is simple and efficient.

[0101] Furthermore, the closeness calculation specifically involves: First, unifying the dimensions of the current sludge granulation state parameter characteristics with the key features of each control strategy in the control strategy library, ensuring that both extract feature values ​​for the same evaluation dimensions such as granulation rate, sludge volume index, and particle size distribution; then, standardizing the two types of feature values ​​to eliminate the influence of dimensional differences; subsequently, using methods such as Euclidean distance closeness and cosine closeness, substituting the standardized feature values ​​of the current state parameters and the standardized feature values ​​of each strategy's key features into the corresponding calculation formula to obtain the closeness value for each comparison; finally, quantifying the similarity degree through the closeness value, the closer the value is to the optimal threshold, such as 1, the higher the matching degree between the current state and the corresponding control strategy, and vice versa.

[0102] Then, the control optimization of sludge granulation state parameters was analyzed using the target sludge process unit control strategy. The control logic and adjustment direction for different granulation states within the target strategy were clarified, and the deviation between the current sludge granulation state parameters and the ideal state parameters corresponding to the target strategy was compared. Based on the deviation analysis results and the operational constraints of the sludge treatment process unit, the types of control parameters requiring adjustment were determined, and the adjustment direction for each parameter was clarified. Subsequently, based on the preset parameter adjustment rules in the target strategy and the allowable adjustment range for the actual operation of the process unit, the specific values ​​of each control parameter were refined to ensure that the adjusted parameters not only meet the requirements of the target strategy but also adapt to the actual operating capacity of the process unit, thus completing the determination of the control parameters for the sludge process unit.

[0103] Finally, the determined control parameters for each sludge process unit are output to the corresponding sludge treatment process unit, driving each unit to operate according to the adjusted parameters. During operation, new sludge granulation state parameters are continuously collected through a multi-modal monitoring sensor array, providing real-time feedback on changes in the current granulation state. The newly collected state parameters are then input back into the aforementioned fuzzy matching optimization stage, repeating strategy matching and parameter optimization analysis. Based on the feedback results, the control parameters are dynamically adjusted, forming a cyclical control process of parameter application - state monitoring - strategy matching - parameter optimization, achieving continuous dynamic regulation of the sludge granulation process.

[0104] By employing proximity calculation to achieve simple and direct fuzzy matching optimization to determine the target control strategy, and combining deviation analysis and process constraints to complete the optimization and analysis of control parameters, a closed-loop control process is constructed, realizing dynamic and precise control of the sludge granulation process and ensuring the stable maintenance of the sludge granulation state.

[0105] In summary, the parameter control method for improving the efficiency of sludge granulation treatment provided in this application has the following technical effects: This application collects historical multimodal process datasets from a multimodal monitoring sensor group, performs standardized processing through a multimodal process data standardization channel, converts the data into a unified format, and combines this with a sludge granulation status assessment channel to evaluate, train, and fit the data, thus constructing a complete sludge granulation assessment channel. This channel assesses the status of sludge treatment multimodal process unit data and outputs sludge granulation status parameters. Then, based on a sludge process unit control strategy library, fuzzy matching optimization and control optimization analysis are performed to determine the sludge process unit control parameters and implement closed-loop control. This improves sludge granulation treatment efficiency and maintains stable system operation, achieving precise closed-loop control of sludge granulation treatment and enhancing its efficiency and stability.

[0106] Example 2, as Figure 2 As shown, based on the same inventive concept as in Embodiment 1 above, this application provides a parameter control system for improving the efficiency of sludge granulation treatment, the system comprising: The sludge treatment process unit set acquisition module 1 is used to extract key process flows from the BIOCOS sludge process system to obtain a sludge treatment process unit set, which consists of a sequence of P tank unit, B tank unit and SU tank unit.

[0107] Multimodal process unit data acquisition module 2 is used to sequentially deploy multimodal monitoring sensor groups on the sludge treatment process unit set, and collect sludge treatment multimodal process unit data through the multimodal monitoring sensor groups.

[0108] The sludge granulation state parameter acquisition module 3 is used to build a sludge granulation assessment channel, and to perform sludge state assessment on the sludge treatment multimodal process unit data based on the sludge granulation assessment channel to obtain sludge granulation state parameters.

[0109] The sludge process unit control parameter acquisition module 4 is used to construct a sludge process unit control strategy library based on the sludge treatment process unit set and the sludge granulation treatment target, perform control optimization analysis on the sludge granulation state parameters based on the sludge process unit control strategy library, determine the sludge process unit control parameters, and perform closed-loop control of sludge granulation through the sludge process unit control parameters.

[0110] Furthermore, the multimodal process unit data acquisition module 2 is used to perform the following steps: Monitoring requirements are analyzed sequentially for each process unit in the sludge treatment process unit set to obtain the monitoring requirement parameters and monitoring parameter boundaries of the sludge process units. Based on the monitoring requirement parameters and monitoring parameter boundaries of the sludge process units, the sensor specification parameters of the sludge process units are determined. The monitoring key point set of the sludge treatment process unit set is extracted, and the coverage analysis of the monitoring key point set is performed to obtain the sensor location parameters of the sludge process units. Based on the sensor specification parameters and sensor location parameters of the sludge process units, a multimodal monitoring sensor group is deployed sequentially on the sludge treatment process unit set.

[0111] Furthermore, the sludge granulation state parameter acquisition module 3 is used to perform the following steps: Historical multimodal process datasets of the multimodal monitoring sensor group are collected; based on the characteristic information of the historical multimodal process datasets, standardized steps are analyzed to construct a multimodal process data standardization processing channel; a sludge granulation state assessment system is defined, and the sludge granulation state assessment system is used to evaluate, train, and fit the historical multimodal process datasets to obtain a granulation state assessment channel; the multimodal process data standardization processing channel and the granulation state assessment channel are cascaded and merged to construct the sludge granulation assessment channel.

[0112] Furthermore, the sludge granulation state parameter acquisition module 3 is used to perform the following steps: The sludge granulation state assessment system is subjected to index extraction to obtain a sludge granulation state assessment index set; each assessment index in the sludge granulation state assessment index set is sequentially associated and mapped with the historical multimodal process dataset to obtain a sludge state index association feature dataset; assessment training and fitting are performed based on the sludge state index association feature dataset to generate a sludge state index assessment branch channel set; the sludge state index assessment branch channel set is combined in parallel to obtain the granulation state assessment channel.

[0113] Furthermore, the sludge granulation state parameter acquisition module 3 is used to perform the following steps: Sludge features are extracted from the historical multimodal process dataset to obtain a static feature set and a dynamic feature set of the sludge process. A cascade analysis is performed based on these two sets to generate a sludge process feature recursion tree. A criticality analysis is conducted on each evaluation index in the sludge granulation state evaluation index set to obtain a set of critical coefficients for the sludge evaluation index. Based on this set of critical coefficients, a hierarchical set of associated features for the sludge evaluation index is determined. Finally, the hierarchical set of associated features for the sludge evaluation index is sequentially mapped to the sludge process feature recursion tree to obtain the sludge state index associated feature dataset.

[0114] Furthermore, the sludge process unit control parameter acquisition module 4 is used to perform the following steps: Extract the sludge granulation task index set to obtain the sludge granulation treatment target; extract the control logic of each process unit in the sludge treatment process unit set to obtain the sludge process unit control logic set; use the historical multimodal process dataset to perform correlation analysis on the sludge granulation task index set and the sludge process unit control logic set to construct a sludge process unit control strategy library.

[0115] Furthermore, the sludge process unit control parameter acquisition module 4 is used to perform the following steps: The sludge granulation task index set and the sludge process unit control logic set are associated and separated with the historical multimodal process dataset to obtain the sludge process unit control logic dataset and the sludge granulation task index dataset. The control data of each process unit in the sludge process unit control logic dataset are associated with the sludge granulation task index dataset to obtain the sludge process unit strategy controller set. The sludge process unit strategy controller set is then optimized and fused with multiple objectives to construct a sludge process unit control strategy library.

[0116] Furthermore, the sludge process unit control parameter acquisition module 4 is used to perform the following steps: Inter-unit influence analysis and control coupling are performed on the sludge process unit strategy controller set to generate a sludge process unit strategy coupled controller; based on the sludge granulation task index set, a sludge granulation treatment objective function is constructed; based on the sludge granulation treatment objective function, multi-objective solution and strategy optimization fusion are performed on the sludge process unit strategy coupled controller to construct the sludge process unit control strategy library.

[0117] Furthermore, the sludge process unit control parameter acquisition module 4 is used to perform the following steps: Based on the sludge process unit control strategy library, fuzzy matching optimization is performed on the sludge granulation state parameters to obtain the target sludge process unit control strategy; the target sludge process unit control strategy is then used to perform control optimization analysis on the sludge granulation state parameters to determine the sludge process unit control parameters.

[0118] The parameter control system for improving the efficiency of sludge granulation treatment provided in this embodiment of the invention can execute the parameter control method for improving the efficiency of sludge granulation treatment provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the method.

[0119] Although this application makes various references to certain modules in the system according to the embodiments of this application, any number of different modules can be used and run on user terminals and / or servers. The various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of each functional unit are only for easy distinction between each other and are not used to limit the scope of protection of this invention.

[0120] The specific embodiments described above do not constitute a limitation on the scope of protection of this application. Those skilled in the art should understand that various modifications, combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the scope of protection of this application. In some cases, the actions or steps described in this application can be performed in a different order than that shown in the embodiments and still achieve the desired results. Furthermore, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

Claims

1. A parameter control method for improving sludge granulation treatment efficiency, characterized by, The method includes: Key process flows of the BIOCOS sludge process system are extracted to obtain a set of sludge treatment process units, which consists of a sequence of P tank units, B tank units and SU tank units. Multimodal monitoring sensor groups are deployed sequentially on the sludge treatment process unit set to collect sludge treatment multimodal process unit data; A sludge granulation assessment channel is established, and the sludge state is assessed based on the sludge granulation assessment channel to obtain sludge granulation state parameters from the multimodal process unit data of the sludge treatment. Based on the sludge treatment process unit set and the sludge granulation treatment target, a sludge process unit control strategy library is constructed. Based on the sludge process unit control strategy library, the sludge granulation state parameters are controlled, optimized, and analyzed to determine the sludge process unit control parameters. The sludge granulation closed-loop control is then performed using the sludge process unit control parameters. The establishment of the sludge granulation assessment channel includes: Collect historical multimodal process data sets from the multimodal monitoring sensor group; Based on the characteristic information of the historical multimodal process dataset, a standardized step parsing is performed to construct a multimodal process data standardization processing channel; Define a sludge granulation state assessment system, and use the sludge granulation state assessment system to evaluate, train and fit the historical multimodal process dataset to obtain the granulation state assessment channel. The multimodal process data standardization processing channel and the granulation status assessment channel are connected in series and merged to build the sludge granulation assessment channel; The method for obtaining the granular state assessment channel includes: The sludge granulation state assessment system is subjected to index extraction to obtain a set of sludge granulation state assessment indicators. Each evaluation index in the sludge granulation state evaluation index set is sequentially associated and mapped with the historical multimodal process dataset to obtain a sludge state index associated feature dataset. Evaluation training and fitting were performed based on the sludge state index associated feature dataset to generate a sludge state index evaluation branch channel set. The sludge condition index assessment branch channel set is combined in parallel to obtain the granulation condition assessment channel. The obtained sludge state index associated feature dataset includes: Sludge features are extracted from the historical multimodal process dataset to obtain a static feature set and a dynamic feature set of sludge processes. A cascaded analysis is performed based on the static feature set and dynamic feature set of the sludge process to generate a sludge process feature recursion tree. A criticality analysis was performed on each evaluation index in the set of sludge granulation status evaluation indicators to obtain a set of criticality coefficients for sludge evaluation indicators. Based on the set of key coefficients of the sludge assessment indicators, determine the set of associated feature levels of the sludge assessment indicators; Based on the hierarchical set of associated features of the sludge assessment index, the sludge process feature recursion tree is sequentially associated with the sludge assessment index to obtain the dataset of associated features of the sludge status index. The construction of the sludge process unit control strategy library includes: Extract and obtain the sludge granulation task index set for the sludge granulation treatment target; The control logic of each process unit in the sludge treatment process unit set is extracted to obtain the sludge process unit control logic set. The historical multimodal process dataset is used to perform correlation analysis on the sludge granulation task index set and the sludge process unit control logic set to construct a sludge process unit control strategy library. The process involves using the historical multimodal process dataset to perform correlation analysis on the sludge granulation task index set and the sludge process unit control logic set, and constructing a sludge process unit control strategy library, including: The sludge granulation task index set and the sludge process unit control logic set are associated and split with the historical multimodal process dataset to obtain the sludge process unit control logic dataset and the sludge granulation task index dataset. The control data of each process unit in the sludge process unit control logic dataset are correlated with the sludge granulation task index dataset to obtain the sludge process unit strategy controller set. The sludge process unit strategy controller set is optimized and fused with multiple objectives to construct a sludge process unit control strategy library; The step of performing multi-objective optimization and fusion on the sludge process unit strategy controller set to construct a sludge process unit control strategy library includes: Perform inter-unit influence analysis and control coupling on the sludge process unit strategy controller set to generate a sludge process unit strategy coupling controller; Based on the set of indicators for sludge granulation, a target function for sludge granulation treatment is constructed. Based on the objective function of sludge granulation treatment, the strategy coupling controller of the sludge process unit is subjected to multi-objective solution and strategy optimization fusion to construct the control strategy library of the sludge process unit.

2. The parameter regulation method for improving sludge granulation treatment efficiency according to claim 1, characterized in that, The deployment of multimodal monitoring sensor groups on the sludge treatment process unit set in sequence includes: Monitoring requirements analysis is performed on each process unit in the sludge treatment process unit set in sequence to obtain the monitoring requirement parameters and monitoring parameter boundaries of the sludge process unit. Based on the monitoring requirements parameters and monitoring parameter boundaries of the sludge process unit, the sensor specifications of the sludge process unit are determined. The monitoring key point set of the sludge treatment process unit set is extracted, and the coverage analysis of the monitoring key point set is performed to obtain the sensor location parameters of the sludge process unit. Based on the sensor specifications and location parameters of the sludge process unit, a multimodal monitoring sensor group is deployed sequentially on the sludge treatment process unit set.

3. The parameter control method for improving the efficiency of sludge granulation treatment as described in claim 1, characterized in that, The determination of the control parameters for the sludge process unit includes: Based on the sludge process unit control strategy library, fuzzy matching optimization is performed on the sludge granulation state parameters to obtain the target sludge process unit control strategy. The control strategy of the target sludge process unit is used to analyze and optimize the sludge granulation state parameters to determine the control parameters of the sludge process unit.

4. A parameter control system for improving the efficiency of sludge granulation treatment, characterized in that, The system is used to implement the parameter control method for improving the efficiency of sludge granulation treatment according to any one of claims 1-3, the system comprising: The sludge treatment process unit set acquisition module is used to extract key process flows from the BIOCOS sludge process system to obtain a sludge treatment process unit set, which consists of a sequence of P tank unit, B tank unit and SU tank unit. The multimodal process unit data acquisition module is used to sequentially deploy multimodal monitoring sensor groups on the sludge treatment process unit set, and collect sludge treatment multimodal process unit data through the multimodal monitoring sensor groups; The sludge granulation state parameter acquisition module is used to build a sludge granulation assessment channel, and to perform sludge state assessment on the sludge treatment multimodal process unit data based on the sludge granulation assessment channel to obtain sludge granulation state parameters. The sludge process unit control parameter acquisition module is used to construct a sludge process unit control strategy library based on the sludge treatment process unit set and the sludge granulation treatment target, perform control optimization analysis on the sludge granulation state parameters based on the sludge process unit control strategy library, determine the sludge process unit control parameters, and perform closed-loop control of sludge granulation through the sludge process unit control parameters.

Images